CN100592325C - A Method of Vehicle Existence Detection Based on Image Texture - Google Patents
A Method of Vehicle Existence Detection Based on Image Texture Download PDFInfo
- Publication number
- CN100592325C CN100592325C CN200710188562A CN200710188562A CN100592325C CN 100592325 C CN100592325 C CN 100592325C CN 200710188562 A CN200710188562 A CN 200710188562A CN 200710188562 A CN200710188562 A CN 200710188562A CN 100592325 C CN100592325 C CN 100592325C
- Authority
- CN
- China
- Prior art keywords
- target
- vehicle
- segment
- row
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 19
- 238000001514 detection method Methods 0.000 title description 10
- 230000011218 segmentation Effects 0.000 claims abstract description 10
- 230000001186 cumulative effect Effects 0.000 claims abstract 4
- 238000004364 calculation method Methods 0.000 claims abstract 3
- 238000009825 accumulation Methods 0.000 claims abstract 2
- 230000001960 triggered effect Effects 0.000 claims description 17
- 238000012545 processing Methods 0.000 claims description 7
- 238000012790 confirmation Methods 0.000 claims 3
- 238000013139 quantization Methods 0.000 abstract 1
- 230000000694 effects Effects 0.000 description 8
- 238000013461 design Methods 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- 230000015572 biosynthetic process Effects 0.000 description 2
- 238000002592 echocardiography Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000005755 formation reaction Methods 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000000638 solvent extraction Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000002604 ultrasonography Methods 0.000 description 2
- 238000004804 winding Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
- Traffic Control Systems (AREA)
Abstract
本发明涉及一种基于图像纹理检测车辆存在的方法，将线阵CCD图像利用小波变换获取一行数据的小波系数，找出其中的局部极大点和极小点，并量化为仅包含1、-1和0的序列，每20行进行一次累加，统计5000行数据中累加和出现的频数最高的数值，作为新的背景纹理，以此每隔5000行更新一次背景，逐行进行二值化计算，获取当前行与前19行数据小波变化量化后的累加和，并将其逐段与背景纹理相比较，如果大于某个阈值，则认为该行中的该段为目标，置为1，否则，则置为0，以此对该行数据进行二值化，在此基础上，逐行进行车辆分割，为每个车道设置若干车辆计数器，逐段判断其所属的车道，并进行车辆信息的修改，以此实现实时车辆分割。
The invention relates to a method for detecting the presence of a vehicle based on image texture. The wavelet coefficients of a row of data are obtained from a line array CCD image by wavelet transform, the local maximum points and minimum points are found out, and quantized to only include 1, - The sequence of 1 and 0 is accumulated every 20 lines, and the value with the highest frequency of accumulation and occurrence in the 5000 lines of data is counted as a new background texture, so as to update the background every 5000 lines and perform binarization calculation line by line , to obtain the cumulative sum of the current row and the first 19 rows of data after wavelet change quantization, and compare it with the background texture segment by segment. If it is greater than a certain threshold, consider the segment in this row as the target and set it to 1, otherwise , then it is set to 0, so as to binarize the row of data. On this basis, vehicle segmentation is performed row by row, and several vehicle counters are set for each lane, and the lane to which it belongs is judged segment by segment, and the vehicle information is Modified to achieve real-time vehicle segmentation.
Description
Technical field
The invention belongs to the Video Detection field, be specifically related to a kind of method that detects the vehicle existence based on image texture.
Technical background
At present, transport information detects method commonly used toroid winding, ultrasound wave/Microwave Measurement, laser/infrared detection and video image detection etc.Toroid winding detecting device cost is low, accuracy of detection is higher, but has the shortcoming of installation and maintenance difficulty.Laser/the infrared detector performance is good, precision is high, but cost an arm and a leg, the equipment complexity.But ultrasound wave/microwave detector all weather operations, vehicle classification more easily realize, but be subjected to easily pedestrian and shelter influence, to detect distance shorter.Video detecting method operation is flexibly directly perceived, simple installation, be easy to safeguard and expense on the low side, therefore have more superiority than other detection methods.
Traditional video detecting method is based on face system of battle formations picture, its background image complexity, be unfavorable for that target is cut apart and feature extraction, and background image is simple relatively in the line array CCD imaging, therefore follow-up target is cut apart, feature extraction and description want simple many, this makes that the job stability of system under all weather conditions is easier to guarantee, and the frame rate of linear array CCD image is higher than the frame rate of area array CCD far away, thereby can realize high-precision test, particularly automobile storage with the detection of car speed.
Line array CCD is imaged as a line of road segment segment face at every turn, in order to guarantee the real-time of testing result, require image data is handled by line, and conventional Video Detection algorithm all is based on a pattern system of battle formations picture, therefore, need design the automobile storage that is applicable to linear array CCD image detection method.
Summary of the invention
The objective of the invention is to, provide a kind of and detect the method that vehicle exists based on image texture, this method can effectively suppress the interference that car light and shade are cut apart vehicle, realizes the accurate location in real time of vehicle.
In order to realize above-mentioned task, the present invention takes following technical solution:
A kind of method that detects the vehicle existence based on image texture, it is characterized in that, this method utilizes wavelet transformation to obtain the wavelet coefficient of data line linear array CCD image, find out a little bigger and minimal point of local pole wherein, and be quantified as and only comprise 1,-1 and 0 sequence, per 20 row carry out one-accumulate, add up the highest numerical value of frequency that adds up in 5000 line data and occur, as new background texture, upgrade a background with this every 5000 row, carry out binaryzation line by line and calculate, after obtaining current line and preceding 19 line data small echos and change quantizing add up and, and it is compared with background texture piecemeal, if greater than certain threshold value, think that then this section in this row is a target, be changed to 1, otherwise, then be changed to 0, with this this line data is carried out binaryzation; And be the binaryzation result that 1 segment calls target phase, binaryzation result is that 0 segment is called dead band;
On this basis, carry out vehicle line by line and cut apart, for each track is provided with the plurality of vehicles counter, judge the track under it piecemeal, the modification of the driving information of going forward side by side realizes that with this real-time vehicle cuts apart.
Automobile storage based on image texture of the present invention is carried out line by line in detection method, and its processing time is less than the acquisition time of data line, therefore can guarantee the real-time that detects, and calculate based on the binaryzation of image texture and can effectively eliminate the interference that car light and shade are cut apart vehicle, realize the accurate location in real time of vehicle.The present invention can directly apply to the linear array CCD image collection site, the vehicle in the image is cut apart in real time, for the traffic parameter statistics provides foundation.
Description of drawings
Fig. 1 is the binaryzation algorithm flow chart that the present invention is based on the pavement texture feature;
Fig. 2 is a vehicle partitioning algorithm process flow diagram of the present invention;
Fig. 3 is treatment effect figure on daytime, and wherein, figure is the original ccd image of gathering a), figure b) be the binaryzation design sketch, figure c) be vehicle segmentation effect figure;
Fig. 4 is treatment effect figure in evening, and wherein, figure is the original ccd image of gathering a), figure b) be the binaryzation design sketch, figure c) be vehicle segmentation effect figure.
Below in conjunction with accompanying drawing content of the present invention is described in further detail.
Embodiment
Method of the present invention utilizes wavelet transformation to obtain the wavelet coefficient of data line linear array CCD image, find out a little bigger and minimal point of local pole wherein, and be quantified as and only comprise 1,-1 and 0 sequence, per 20 row carry out one-accumulate, add up the highest numerical value of frequency that adds up in 5000 line data and occur, as new background texture, upgrade a background with this every 5000 row, carrying out binaryzation line by line calculates, after obtaining current line and preceding 19 line data small echos and change quantizing add up and, and it is compared with background texture piecemeal, if greater than certain threshold value, think that then this section in this row is a target, be changed to 1, otherwise, then be changed to 0, with this this line data is carried out binaryzation; And be the binaryzation result that 1 segment calls target phase, binaryzation result is that 0 segment is called dead band;
On this basis, carry out vehicle line by line and cut apart, for each track is provided with the plurality of vehicles counter, judge the track under it piecemeal, the modification of the driving information of going forward side by side realizes that with this real-time vehicle cuts apart.
Referring to Fig. 1, as follows based on the concrete implementation step of the binaryzation algorithm of image texture characteristic:
1) initialization of variable.Find background indicia bFindBack=false; Current every trade AL=0, background line number counter nBlock=0,20 line data add up by row and nSum[]=0, add up the frequency aBlockInfo[that each segment data occurs] []=0, binaryzation parameter alpha=0.35, β=2.0;
2) from internal memory, read data line, put AL=AL+1, nBlock=nBlock+1;
3) this line data is carried out three grades of wavelet transformations, takes out wavelet coefficient, be stored in array pDWTData[] in;
4) with array pDWTData[] only be converted into and comprise-1,0,1 sequence, be stored in AR[] in.Wherein, with pDWTData[] in local pole be labeled as 1 a little louder, local minimum point is labeled as-1, other points are labeled as 0;
5) if step 2 is changeed in AL＜20);
6), and be stored in array nSum[by the row AR value of up-to-date 20 line data that adds up] in;
7) getting window size is 16, to nSum[] in data handle by window, in window ranges, get the value of a maximal value, and be stored in array nSumAvg[as all pixels in this window] in, promptly data line is divided into 64 sections;
8) if bFindBack=false changes step 11);
9) according to background aBack[], carry out binaryzation by following condition;
if(nSumAvg[j]＜＝α*aBack[j]&&nSumAvg[j]＞＝β*aBack[j])
The pairing pixel two-value of this row j segment data turns to 1;
else
The pairing pixel two-value of this row j segment data turns to 0;
10) the binaryzation result to data line further handles, if dead band length≤2 between two target phases, then the pixel of this dead band correspondence is filled to 1, if the length of previous target phase≤2, and distance 〉=7 of the back target phase of its distance then are changed to 0 to the pixel of previous target phase correspondence;
11) frequency that occurs of the AR value of up-to-date 20 line data of statistics, and be stored in array aBlockInfo[] in [], this array first dimension is represented the value of AR, and second ties up and represents segment number;
12) if step 2 is changeed in nBlock＜5000);
13) according to aBlockInfo[] statistical information of [], find out each section and the maximum numerical value of frequency occurs as new background, and be stored in aBack[] in, put nBlock=0, revise background indicia bFindBack=true, and put aBlockInfo[] []=0;
14) if receive command for stopping, then stop data processing, otherwise change step 2).
Be the convenient vehicle partitioning algorithm of describing, the line array CCD visual field is divided into three logic tracks (corresponding to two monitored physics tracks) according to the close shot camera field of view, and each track comprises some information: pre-trigger sign, confirm to trigger sign, there are counter in null counter, target, target left margin, target right margin, target begin column number, target end line number, trigger the definite line number of point midway, target level projection counter, triggering in advance.
Referring to Fig. 2, the concrete implementation step that vehicle is cut apart is as follows:
1) lane information initialization.Pre-triggering sign with each track, confirm to trigger sign and be changed to false, with the null counter, the target left margin, the target right margin, target begin column number, target end line number, target level projection counter, the definite line number that triggers is changed to 0, objective definition segment length threshold value n1, the level thresholds n2 that merges vehicle, the horizontal projection of determining headstock adds up and threshold value n3, confirm the target width threshold value n4 of triggering, confirm the target length threshold value n5 of triggering, the null that the affirmation target finishes is counted n6, target maximum length n7, the minimum length n8 of target determines it is not that the horizontal projection of headstock adds up and threshold value n9;
2) read the binaryzation result of up-to-date data line;
3) reference position of the next target phase in location;
4) change step 5) if continuous target phase number surpasses n1, otherwise change step 3);
5) judge whether this continuous target phase can merge to the track of having triggered, if can merge, then changes step 6), otherwise change step 7);
Judge track under it according to the mid point of successive objective section,, need judge whether this continuous target phase belongs to the vehicle of triggering in advance in order to prevent false triggering.The objective definition section to the distance that triggers the track in advance is: if this continuous target phase has common factor with the vehicle horizontal extent that triggers the track, think that then this target phase belongs to the vehicle that has triggered in this track, its distance is 0; Otherwise, the minor increment on border, the left and right sides of vehicle in track has been triggered as its distance in the border, the left and right sides of getting continuous target phase distance respectively, if this distance surpasses threshold value n2, think that then this target phase does not belong to the vehicle that has triggered in this track, otherwise think that this target phase belongs to the vehicle that has triggered in this track;
6) revise the boundary information of this track vehicle.The number that target level projection counter is added up target phase piecemeal, the end line of modifying target number, if the border, the left and right sides of this target phase has exceeded the left and right sides bounds in this track, then step 8) is changeed on the border, the left and right sides of modifying target;
7) newly-built information of vehicles on the track under the successive objective section is put pre-triggering and is masked as true, opens target level projection counter, and border, the left and right sides, begin column number, the end line number of target are set;
8) judge whether to handle all segments of this line data,, change step 9), otherwise change step 3) if dispose;
9) handle information of vehicles by the track;
10) if there is target phase to occur on this track, the null counter puts 0, changes step 11), otherwise changes step 14);
11) if this track is not also confirmed to trigger, change step 12), otherwise change step 15);
12) judge whether to satisfy the condition of confirming triggering,, change step 13), otherwise change step 16) if satisfy.Confirm that trigger condition is: the horizontal projection of headstock information adds up and surpasses threshold value n3, and the width of target surpasses threshold value n4, and the length of target is above threshold value n5;
13) put the affirmation triggering and be masked as true, the definite line number of triggering is changed to the beginning line number of target, changes step 19);
14) this track null counter adds 1, if null length surpasses threshold value n6, thinks that then target finishes, and changes step 18), otherwise change step 19);
15) judge whether target length surpasses threshold value n7,, think that then target finishes, and changes step 18 if surpass), otherwise change step 19);
16) judge whether to satisfy the condition that cancellation triggers,, then change step 17 if satisfy), otherwise change step 19).The cancellation trigger condition is: target length surpasses threshold value n8, and perhaps the target length horizontal projection that surpasses n5 and headstock information adds up and less than threshold value n9;
17) cancellation triggers, put pre-triggering sign, confirm that the triggering sign is changed to false, the definite line number of null counter, target left margin, target right margin, target begin column number, target end line number, target level projection counter, triggering is changed to 0, changes step 19);
18) carrying out vehicle cuts apart.Determine the border, the left and right sides of vehicle according to target level projection counter, border, front and back according to the begin column of vehicle number and end line number definite vehicle, sign vehicle scope, and put pre-triggering sign, confirm to trigger sign and be changed to false, the definite line number of null counter, target left margin, target right margin, target begin column number, target end line number, target level projection counter, triggering is put 0;
19) judge whether to handle all tracks,, change step 20 if dispose), otherwise change step 9);
20) if receive command for stopping, then stop data processing, otherwise change step 2).
It below is the specific embodiment that the inventor provides.
Embodiment 1:
With reference to Fig. 3 a-c) shown in, the linear array CCD image that figure a) gathers for daytime is because there is vehicle shadow in the influence of the sun in the image.Employing based on the binaryzation algorithm process effect of image texture characteristic as figure b) shown in, the interference of having removed shade, figure c) be the vehicle segmentation effect, as seen the present invention has effectively suppressed the influence that shade is cut apart vehicle.
Embodiment 2:
With reference to Fig. 4 a-c) shown in, the linear array CCD image that figure a) gathers for evening is because the influence of car light and light filling exists in the image and not only has car light, and has shade.Employing based on the binaryzation algorithm process effect of image texture characteristic as figure c) shown in, the interference of having removed light and shade, figure c) be the vehicle segmentation effect, as seen the present invention has effectively suppressed the influence that shade and car light are cut apart vehicle.
Claims (1)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200710188562A CN100592325C (en) | 2007-12-11 | 2007-12-11 | A Method of Vehicle Existence Detection Based on Image Texture |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200710188562A CN100592325C (en) | 2007-12-11 | 2007-12-11 | A Method of Vehicle Existence Detection Based on Image Texture |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101231699A CN101231699A (en) | 2008-07-30 |
| CN100592325C true CN100592325C (en) | 2010-02-24 |
Family
ID=39898160
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN200710188562A Expired - Fee Related CN100592325C (en) | 2007-12-11 | 2007-12-11 | A Method of Vehicle Existence Detection Based on Image Texture |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100592325C (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103903434B (en) * | 2012-12-28 | 2015-11-04 | 重庆凯泽科技有限公司 | Based on the intelligent transportation system of image procossing |
| CN107240101B (en) * | 2017-04-13 | 2021-01-29 | 桂林优利特医疗电子有限公司 | Target area detection method and device, and image segmentation method and device |
-
2007
- 2007-12-11 CN CN200710188562A patent/CN100592325C/en not_active Expired - Fee Related
Non-Patent Citations (7)
| Title |
|---|
| An Efficient Method of Locating Vehicle License Plate. Zhigang Xu, Honglei Zhu.Third International Conference on Natural Computation,Vol.2 . 2007 |
| An Efficient Method of Locating Vehicle License Plate. Zhigang Xu,Honglei Zhu.Third International Conference on Natural Computation,Vol.2. 2007 * |
| 一种新的汽车牌照识别的图像增强算法. 宋焕生，赵祥模，王养利.长安大学学报（自然科学版），第26卷第4期. 2006 |
| 基于小波分解的车辆视频检测算法. 陈晓梅，黄宏涛.华东交通大学学报，第23卷第2期. 2006 |
| 基于小波分解的车辆视频检测算法. 陈晓梅，黄宏涛.华东交通大学学报，第23卷第2期. 2006 * |
| 视频交通监控系统中背景提取算法. 郭永涛，宋焕生，贺昱曜.电视技术，第5期. 2006 |
| 视频交通监控系统中背景提取算法. 郭永涛，宋焕生，贺昱曜.电视技术，第5期. 2006 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101231699A (en) | 2008-07-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhu et al. | VISATRAM: A real-time vision system for automatic traffic monitoring | |
| CN101514993B (en) | Vehicle speed measurement device based on linear array CCD camera | |
| CN100545867C (en) | Aerial shooting traffic video frequency vehicle rapid checking method | |
| CN107301776A (en) | Track road conditions processing and dissemination method based on video detection technology | |
| CN102426785B (en) | Traffic flow information perception method and system based on contour and local feature points | |
| CN101364347A (en) | A video-based detection method for vehicle control delays at intersections | |
| CN104063882B (en) | Vehicle video speed measuring method based on binocular camera | |
| CN101872546A (en) | A fast detection method for cross-border vehicles based on video | |
| CN110298300A (en) | A method of detection vehicle violation crimping | |
| CN110718061A (en) | Traffic intersection traffic flow statistics method, device, storage medium and electronic device | |
| CN114998317B (en) | Lens occlusion detection method and device, camera device and storage medium | |
| Telagarapu et al. | A novel traffic-tracking system using morphological and Blob analysis | |
| CN100592325C (en) | A Method of Vehicle Existence Detection Based on Image Texture | |
| CN208271388U (en) | A kind of volume of traffic acquisition system based on video monitoring | |
| Mossi et al. | Real-time traffic analysis at night-time | |
| Hai-Feng et al. | Vehicle abnormal behavior detection system based on video | |
| CN1350941A (en) | Method and device for moving vehicle image tracking | |
| Janda et al. | A road edge detection approach for marked and unmarked lanes based on video and radar | |
| Al Okaishi et al. | Vehicular queue length measurement based on edge detection and vehicle feature extraction | |
| CN101216892A (en) | A Method for Recognizing the Existence of Vehicles Based on the Gray Scale Features of Line Array CCD Camera Sequence Images | |
| CN113160299B (en) | Vehicle video speed measurement method based on Kalman filtering and computer readable storage medium | |
| CN110516549B (en) | Traffic flow parameter obtaining method based on binary time domain diagram | |
| Al-Kadi et al. | Road scene analysis for determination of road traffic density | |
| Trivedi et al. | Vehicle counting module design in small scale for traffic management in smart city | |
| Li et al. | An Effective Algorithm for Video‐Based Parking and Drop Event Detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20100224 Termination date: 20121211 |